Back to the Top
Dear All,
What are the standard BQL set of rules used in PK or PD modeling (Using NONMEM / PHOENIX) ?
With best regards,
Raghav,
Back to the Top
The following message was posted to: PharmPK
Raghav,
There are no standard BQL rules in PK-PD modeling: all depend on data. If BQL is the last in the series of the terminal-phase measurements and thus has no information value, it can be removed from the dataset. If the large fraction of the data is informative BQLs (that do contribute to the information content of the dataset), BQLs need to be preserved and used for modeling. How to use them is the question that has been discussed numerous times, see
J Pharmacokinet Pharmacodyn. 2008 Aug;35(4):401-21. Epub 2008 Aug 7.
Likelihood based approaches to handling data below the quantification limit using NONMEM VI. Ahn JE, Karlsson MO, Dunne A, Ludden TM.
Erratum in:
J Pharmacokinet Pharmacodyn. 2010 Jun;37(3):305-8.
Leonid
Back to the Top
The following message was posted to: PharmPK
Dear Raghav:
There is no need to censor low data. There is no BLQ. This has been
discussed ad nauseam on PharmPK. CV% is an obsolete measure of assay
precision, as, of course, when the measurement gets low, the CV goes up.
Find a measure of precision that all the rest of the world uses except the
lab community, the reciprocal of the assay variance for each measurement.
There is really no reason at all to censor low data. There is no BLQ in is
an illusion brought about by using CV%.
For a start, you might look at
Jelliffe RW, Schumitzky A, Van Guilder M, Liu M, Hu L, Maire P, Gomis P,
Barbaut X, and Tahani B: Individualizing Drug Dosage Regimens: Roles of
Population Pharmacokinetic and Dynamic Models, Bayesian Fitting, and
Adaptive Control. Therapeutic Drug Monitoring, 15: 380-393, 1993.
Very best regards,
Roger Jelliffe
Back to the Top
I would tend to agree with Roger re the %CV. However, precision is only one parameter, the accuracy of a measure is the other and both %CV and %accuracy measures determine the LLOQ or BLQ points, as they do the ULOQ points. It matters little if the precision error is small but the accuracy error is great. Would anyone use values associated with a relative error of 80% but a %CV of 5%? Is there another way around this? With the current definitions I believe not.
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "BQL set of rules" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)